From 005a3c4b8484917358de8e6267d5d9bf7fb24aa7 Mon Sep 17 00:00:00 2001
From: "juvenal.diaz" <juvenal.diaz@oracle.com>
Date: Wed, 22 Jul 2026 23:10:11 -0600
Subject: [PATCH] Make helm do not block

---
 bootstrap/platform/main.tf | 51 +++++++++++++++++++++++++++++++++++---
 1 file changed, 48 insertions(+), 3 deletions(-)

diff --git a/bootstrap/platform/main.tf b/bootstrap/platform/main.tf
index df96930..072e5c1 100644
--- a/bootstrap/platform/main.tf
+++ b/bootstrap/platform/main.tf
@@ -1770,8 +1770,10 @@ resource "helm_release" "prometheus_stack" {
   version          = var.observability.prometheus.version
   namespace        = var.observability.namespace
   create_namespace = false
-  timeout          = 1200
-  wait             = true
+  # Readiness is checked by the following resource so a blocked PVC or pod
+  # produces actionable Kubernetes diagnostics instead of a provider timeout.
+  timeout          = 600
+  wait             = false
   cleanup_on_fail  = true
 
   values = [
@@ -1973,6 +1975,49 @@ resource "helm_release" "prometheus_stack" {
   ]
 }
 
+resource "null_resource" "prometheus_stack_readiness" {
+  depends_on = [helm_release.prometheus_stack]
+
+  triggers = {
+    kubeconfig_path = var.kubeconfig_path
+    namespace       = var.observability.namespace
+    release_version = var.observability.prometheus.version
+  }
+
+  provisioner "local-exec" {
+    interpreter = ["/bin/bash", "-lc"]
+    command     = <<EOT
+set -euo pipefail
+
+kubeconfig="${self.triggers.kubeconfig_path}"
+namespace="${self.triggers.namespace}"
+selector="app.kubernetes.io/instance=prometheus-stack"
+timeout_seconds=600
+elapsed=0
+
+dump_prometheus_debug() {
+  echo "Prometheus stack readiness diagnostics:" >&2
+  kubectl --kubeconfig "$kubeconfig" -n "$namespace" get pods,sts,deploy,pvc -l "$selector" -o wide >&2 || true
+  kubectl --kubeconfig "$kubeconfig" -n "$namespace" get events --sort-by=.lastTimestamp | tail -n 80 >&2 || true
+}
+
+trap dump_prometheus_debug ERR
+
+until kubectl --kubeconfig "$kubeconfig" -n "$namespace" get pods -l "$selector" -o jsonpath='{range .items[*]}{.metadata.name}{"\t"}{.status.phase}{"\t"}{range .status.conditions[*]}{.type}={.status}{" "}{end}{"\n"}{end}' 2>/dev/null |
+  awk 'NF > 0 { count++ } ($2 == "Succeeded") || ($2 == "Running" && $0 ~ /Ready=True/) { next } { failed = 1 } END { exit !(count > 0 && !failed) }'; do
+  if [ "$elapsed" -ge "$timeout_seconds" ]; then
+    echo "Timed out waiting for Prometheus stack workloads after $${timeout_seconds}s." >&2
+    exit 1
+  fi
+  sleep 10
+  elapsed=$((elapsed + 10))
+done
+
+echo "Prometheus stack workloads are ready."
+EOT
+  }
+}
+
 resource "kubernetes_config_map" "grafana_dashboards" {
   for_each = local.grafana_dashboard_files
 
@@ -1991,7 +2036,7 @@ resource "kubernetes_config_map" "grafana_dashboards" {
     (each.key) = file("${path.module}/grafana-dashboards/${each.key}")
   }
 
-  depends_on = [helm_release.prometheus_stack]
+  depends_on = [null_resource.prometheus_stack_readiness]
 }
 
 resource "helm_release" "extra_tools" {